The purpose of sub-project 4 is to complete a longitudinal comparison of imitation based and conversational recast interventions in 72 children with specific language impairment. The issue examined in this study is whether there are optimal developmental language levels for delivering focused "conversational recast" responses to child initiations and/or for structured imitation presentation of language targets. From a transactional perspective, the ability of a child with specific language impairment to learn from these different types of adult responses may shift as the child's language level changes. Our previous studies of the effects of conversational recast and imitation interventions (Camarata, Nelson and Camarata, 1994; Nelson et al. 1996) were limited because the treatments were provided to SLI children selected from a relatively narrow developmental language level (Brown Stages IV/V). This sub-project will extend our previous research on conversational recast to more diverse language levels and will measure generalized impacts of target acquisition on language outcomes (spontaneous language samples at home and in the clinic settings). Additionally, in terms of the overall conceptualization of the program project (see overview), this sub-project will address the following issues: Examining the potential association between language growth and pre-intervention measures (e.g., developmental language level at intake, results of standardized tests, including pre-intervention levels of comprehension); and evaluating the nature and extend of practice required for target acquisition, and for generalization and maintenance.